PCA (Data Shrinker)

→ חזרה למאמר

PCA makes a huge table smaller while trying to keep the most important pattern.

What it does

PCA makes a huge table smaller while trying to keep the most important pattern.

Columns: The kinds of facts in a table • Spread: How different the numbers are

How it works

  • Look for columns that say almost the same thing.
  • Find the strongest direction in the data.
  • Keep fewer columns so the computer has less to carry.

Uses and applications

  • A phone can shrink image data before another AI studies it.
  • A science tool can simplify a giant spreadsheet.
  • A dashboard can show a big pattern on a simple chart.

In short

PCA is like squishing a 3D object into a flat shadow.

Related methods

  • K-Means: Often applied after PCA reduces the number of features.
  • VAE (Autoencoders): Also learns a compact code, using a neural network.
  • Word2Vec: Represents items as points in a lower-dimensional space.

Back to Patterns, anomalies and recommendations · All topics

The interactive demonstration and the Python walkthrough for this topic need JavaScript.